Title: ESSnet on the use of administrative and accounts data in business statistics
1ESSnet on the use of administrative and accounts
data in business statistics Development of
Quality Indicators (WP6)
John-Mark Frost (ONS, UK), Humberto Pereira (INE,
PT), Sofia Rodrigues (INE, PT), Ana Chumbau (INE,
PT), Jorge Mendes (INE, PT) and Sarah Green
(ONS, UK), Q2010, 4th May 2010
2Overview
- Work package (WP) partners
- Aims of the WP
- Why this work is important
- Work done so far
- Next steps
3WP partners
4Aims of the WP (2009 2013)
- To collect and analyse information on existing
methods used in NSIs for quality assessment when
administrative data are used. - To develop quantitative quality indicator(s) for
business statistics produced using administrative
data, and - To develop qualitative indicators to complement
the quantitative one(s).
5Why this work is important
- Increasing use of administrative data in business
statistics - Dimensions of quality apply but
- - Quality reporting is not entirely the same
- - CVs cannot be used when solely using
administrative data - Best practice in other areas will depend on
appropriate and effective measures of quality
6Work done so far
- Focussed on
- collecting and analysing information on existing
methods used in NSIs for quality assessment when
administrative data are used.
7Phase 1 Methodology
- Developed Questionnaire
- Use of administrative data in business statistics
- Use of quality checks
- Circulated to 34 NSIs
- 27 Member States
- 4 EFTA
- 3 Non-European
- 90 response rate
8Phase 1 Results (1)
- Administrative data used extensively in business
statistics - Quality considered an important issue
- Lots of generic checks conducted during the
process of statistics production
9Phase 1 Results (2)
- But
- checks are not necessarily formal and very few
are published as quality indicators - only half the NSIs indicated that they produce
any kind of quantitative quality measure
10Phase 2 Methodology
- Identified 16 NSIs that showed the most
experience in the area of quality indicators - Sent more extensive questionnaires, specifically
asking about quantitative quality indicators in
the areas of SBS, STS, Business Registers and
Prodcom - 100 response rate
11Phase 2 Results
- Consistent with Phase 1
- NSIs check quality
- Checks are generally made as part of the
statistical production process - But the checks are not necessarily produced on
a regular or formal basis.
12Phase 3 Methodology
- Identified 7 more experienced NSIs and
requested to meet with them - Engaged in face-to-face interviews with relevant
staff within the NSIs to - - Better understand their use of administrative
data in business statistics - - Gain clarity on their responses in Phase 2
- Get more detailed information on their use of
quality indicators
13Phase 3 Results (1)
- NSIs engaged in similar quality checks
- Accuracy
- e.g. of units with correct activity code
- Coverage
- e.g. comparison of units included in
administrative source with units in the BR to
estimate under / over-coverage - Missing data/non-response,
- e.g. of turnover accumulated at publication of
first estimate
14Phase 3 Results (2)
- Revisions
- e.g. differences between first and final
estimates - Matching (more relevant for NSIs without unique
identifiers) - e.g. of matched units from both sources
- Coefficients of Variation (when combining
administrative and survey data) - e.g. using the jack-knife method
- However, very few of these checks were produced
as formal, quality indicators
15Summary of work so far
- Administrative data are widely used in business
statistics - Quality is seen as important
- Various checks are conducted during the
statistical production process but - they are not necessarily formal or regular
- they are rarely published
- On the whole, NSIs do not produce quality
indicators in the same way as when using survey
data - Development of list of quality indicators
welcomed by NSIs
16The next steps
- Build on the results of the stock-take research
- Further develop the list of quality indicators
- including user testing
- Investigate composite quality indicators
- Throughout, ensure that we
- adopt a pragmatic approach
- develop a user-friendly list of indicators
- consider the limitations on NSIs (resource and
data availability)
17Some areas for consideration
- Indicators that apply to both survey and
administrative data - Not all NSIs will have access to the same level
or type of information (either administrative or
process related) - Qualitative as well as quantitative indicators
18Thank you for listening